Biodiversity loss, ecosystem degradation, and habitat destruction are increasingly linked to human-driven changes in land use, including urbanisation, agriculture, and the exploitation of natural resources (European Parliament, 2025; Jaureguiberry et al., 2022). In response, governments across Europe — including the EU — have introduced ambitious environmental strategies such as the EU Biodiversity Strategy for 2030 (European Parliament, 2025) and the 30x30 target (Markwick, 2023), which aims to protect 30% of land and sea by the year 2030.
Ecological restoration plays a vital role in addressing these challenges. Rather than simply returning ecosystems to a previous state, modern approaches focus on restoring ecological processes and enhancing ecosystem resilience (Hicks, 2023).
RestREco (Restoring Resilient Ecosystems) is a NERC-funded research project that adopts a resilience-based perspective on ecological restoration. The initiative brings together researchers from:
Using a natural experiment design, RestREco studies a network of 133 ecological restoration sites across England and Scotland. The project aims to identify key drivers of ecosystem development, such as:
The goal is to understand how these factors influence ecosystem complexity, function, and resilience to future pressures (RestREco, 2024).
As part of the RestREco initiative, the Dig Deeper study focused on how the age of restoration, establishment type, and site management affect soil microbial communities, specifically bacteria and fungi.
To explore this, high-throughput sequencing was conducted on:
The analysis focused on three main aspects:
These microbial assessments complement broader ecosystem-level measurements within the RestREco project, including vegetation, invertebrates, and ecosystem functions such as litter decomposition, pollination services, and soil thermodynamic efficiency.
The following sections describe the sampling design, metadata structure, and the processing pipeline used to characterise microbial communities.
This study investigates how grassland site age, establishment method, management practices, and soil pH influence the diversity, taxonomic composition, and functional profiles of bacterial and fungal communities during restoration. It also explores the interactions between bacteria and fungi, focusing on potential correlations between microbial taxa, functional pathways, and fungal guilds.
The following hypotheses were formulated:
A total of 330 soil samples were collected in 66 sites of England for each marker (5 per site).
Sample Zone - Based on GPS coordinates
| Metric | 16S | ITS |
|---|---|---|
| Microbial group | Bacteria | Fungi |
| Region sampled | England | England |
| Number of sites | 66 | 66 |
| Samples per site | 5 | 5 |
| Total samples | 330 | 330 |
| Average reads per sample | ~65,000 | ~65,000 |
| Read count range | 30,000–85,000 | 10,000–90,000 |
Sampling Summary
Each sample collected was accompanied by metadata capturing key environmental and management variables. These contextual factors were essential for interpreting variation in microbial diversity.
| Variable | Description |
|---|---|
| Site | Name of the sampling site |
| Plot number | Subdivision of each site (usually 5 plots per site) |
| CU Code | Unique code for each sample |
| Year_est | Year of establishment of the site |
| Age | Site age (ranging from 1 year to over 100 years ) |
| Latitude/Longitude | GPS coordinates of the sample |
| Establishment | Restoration type or land management |
| pH | Soil pH value at the time of sampling |
| EC | Electrical conductivity of the soil |
| Cutting | Whether the site is cut (1 = Yes, 0 = No) |
| Cattle | Presence of cattle grazing (1 = Yes, 0 = No) |
| Sheep | Presence of sheep grazing (1 = Yes, 0 = No) |
| Plough | Whether the soil has been ploughed (1 = Yes, 0 = No) |
Metadata Summary
A total of 330 soil samples were collected across 66 sites (five per site) by the RestREco team. After sieving (2 mm), all samples were frozen until DNA extraction, which was performed using the QIAGEN PowerSoil kit. Amplicon sequencing targeted two regions: the V4–V5 region of the bacterial 16S rRNA gene and the ITS1 region of the fungal ITS gene. Sequencing was carried out by Novagene using Illumina paired-end reads (2×250 bp), resulting in 660 FASTQ files per marker (forward and reverse reads per sample). Metadata accompanying each sample included site characteristics (location, year of establishment, method), management practices (cutting, ploughing, grazing), and soil physicochemical properties.
Quality control of 16S reads was performed using FastQC
v0.12.1 and summarised with MultiQC v1.14. A
custom Bash script (S02_qc.sh) ensured correct pairing of
reads and removed five incomplete samples. Sequences were then imported
into QIIME2 v2022.10 using the
PairedEndFastqManifestPhred33V2 format, and primers were
removed with CutAdapt v4.4. Read quality was visualised
using qiime demux summarize.
Denoising and read pairing were performed using the DADA2 plugin (QIIME2 v2024.2), generating ASV feature tables, representative sequences, and denoising statistics. Low-abundance ASVs (<10 reads) and singletons were filtered out. Representative sequences were aligned using MAFFT and used to build phylogenetic trees with FastTree.
The ASV table was aggregated at the site level using a median-ceiling method, and rarefaction analysis was used to determine a sampling depth of 9473 reads/sample. Alpha (Observed Features, Shannon, Evenness, Faith’s PD) and beta diversity (UniFrac, Jaccard, Bray-Curtis) metrics were calculated using this rarefied table and the rooted tree. Group comparisons (e.g., by establishment, livestock, pH) were assessed using Kruskal-Wallis (alpha) and PERMANOVA (beta).
Taxonomy was assigned using a Naive Bayes classifier trained on Greengenes 13_8 (515F/806R). Taxonomic barplots were created from feature tables grouped by site. Tables were then collapsed at genus and family levels for ANCOM differential abundance testing. Pseudocounts were added where required.
Predicted functional profiles were generated using PICRUSt2 to infer KEGG Orthology (KO), EC numbers, and pathways. High-abundance pathways were retained (≥600,000 reads in ≥60 samples), and visualised with heatmaps across metadata variables. Functional differential abundance was assessed using ANCOM.
Functional alpha and beta diversity metrics were also computed, and significance was tested per metadata variable, using Kruskal-Wallis and PERMANOVA, respectively.
Fungal ITS1 reads were processed using ITSxpress
v2.1.4, which performed adapter trimming and retained fungal
ITS1 regions. Quality was assessed with FastQC and
summarised using MultiQC v1.28
(F01_itsexpress.sh). Trimmed reads were imported into
QIIME2 v2024.10, and denoised using
DADA2 with no truncation.
ASV tables were aggregated by site and taxonomic assignment was performed using a pre-trained classifier based on the UNITE v10 database. Diversity analyses (Observed Features, Shannon, Evenness, Jaccard, Bray-Curtis) were conducted using a rarefied depth of 30,000 reads/sample, and visualised using PCoA and Emperor plots. Significance testing followed the same approach as for 16S (Kruskal-Wallis and PERMANOVA).
Taxonomic tables were collapsed at the genus level for
ANCOM analyses. Ecological functional guilds were
predicted using FUNGuild v1.1, based on collapsed
species-level data formatted with custom Python scripts. Functional
diversity metrics were then computed at the guild
level, and visualised using core-metrics tools at a rarefied
depth of 11,000 reads/sample. Significance was assessed
using Kruskal-Wallis and PERMANOVA
(F09 and F10 scripts).
To explore inter-kingdom interactions, correlation
analyses were conducted between bacterial and fungal taxa and
between predicted bacterial pathways (PICRUSt2) and
fungal functional guilds (FUNGuild). Using R
v4.4.2 and the psych::corr.test function,
Pearson correlations were calculated at the genus
level. Results were adjusted using Holm-Bonferroni
correction. Only strong and significant correlations (e.g., |r| >
0.7, adjusted p ≤ 0.05) were retained after filtering. Additional
filters removed taxa/guilds with low overall abundance.
For functional correlations, similar thresholds were applied, but due to limited signal strength, analyses were interpreted conservatively and filtered results were reported alongside raw values when relevant.
Workflow (16S)
You can explore the full MultiQC report by clicking the image below:
After denoising, quality control (QC) metrics provide essential insights into the effectiveness of the data processing steps and the overall quality of the resulting feature table. This summary table presents key statistics for each sample, including the number of input reads, filtered reads, and final feature counts after denoising. These metrics help assess sequencing success, identify potential outliers, and ensure that sufficient data remain for robust downstream analyses. Samples with unusually low read counts or feature richness may need to be excluded or interpreted with caution.
Here is a link to the statistics after denoising to view it on QIIME2 (16S) : Statitics after denoising (16S)
Statitics after denoising (16S)
Here is a link to the statistics after denoising to view it on QIIME2 (ITS) : Statitics after denoising (16S)
Statitics after denoising (ITS)
These Quality Control (QC) plots were generated after trimming the sequencing reads. They provide a visual summary of the base quality scores, read length distributions, and other metrics, helping to assess whether the trimming step successfully removed low-quality regions and adapter contamination. Consistently high-quality reads across samples are essential for reliable downstream analysis.
Rarefaction curves provide a visual tool to assess sequencing depth and compare species richness between samples. In this study, rarefaction curves were generated separately for bacterial (16S rRNA gene) and fungal (ITS) communities, using the number of observed features—i.e., unique ASVs—as a proxy for richness.
For both bacteria and fungi, the shape of each curve indicates whether sequencing depth was sufficient to capture most of the diversity in a given sample. Curves that level off suggest that a representative portion of the community has been sampled, whereas rising curves indicate that additional sequencing could reveal further diversity.
This step is crucial to ensure that downstream diversity analyses are not biased by unequal sampling effort.
Here is a link to the Rarefiction plots for more flexibility on QIIME2: Rarefiction plots (16S)
Rarefaction Curves of Observed Features by 'Establishment' (16S)
Here is a link to the Rarefiction plots for more flexibility on QIIME2: Rarefiction plots (ITS)
Rarefaction Curves of Observed Features by 'OS_location' (ITS)
Summary
Microbial taxonomic diversity revealed clear responses to restoration
strategies. For bacteria, establishment method was the strongest driver
of alpha and beta diversity, followed by pH and grazing. For fungi, pH
and age were more influential. NR sites generally showed lower bacterial
diversity and functional richness than GH and SM.
Alpha diversity refers to the variety of organisms within a particular sample or environment. It reflects both richness—the number of distinct taxa—and evenness—how evenly individuals are distributed among those taxa. One of the most widely used measures for assessing alpha diversity is the Shannon index.
The Shannon index takes into account not only the number of species present, but also how evenly their abundances are distributed. A higher Shannon value generally indicates a more diverse and ecologically balanced community.
Another important metric is Faith’s Phylogenetic Diversity (Faith PD), which measures the total branch length of the phylogenetic tree that spans the species in a sample. Unlike the Shannon index, Faith PD incorporates evolutionary relationships, providing a phylogenetic perspective on diversity.
We also include Pielou’s Evenness index, which specifically quantifies how equally individual organisms are distributed across taxa. While Shannon integrates both richness and evenness, this metric isolates the evenness component, providing a complementary view of diversity patterns.
To allow interactive exploration of alpha diversity metrics across
different environmental variables, we implemented a drop-down menu that
dynamically displays the corresponding plots. Some variables, such as
pH category, are only present in the ITS dataset, while
others, like Year group, are specific to the 16S dataset.
Internally, variables are mapped to their dataset-specific equivalents
where needed (e.g. Age group in 16S becomes
Age category in ITS). It is important to note, however,
that these variables are not always directly comparable: for instance,
Age group (16S) divides sites into multiple discrete
intervals based on restoration age, while Age category
(ITS) is a binary classification based on whether a site is above or
below the median age. Despite these differences, the interface ensures
that only available and relevant plots are shown for each selection.
Summary
Bacterial alpha diversity (Shannon, Faith PD, Evenness) significantly
varied with establishment method. GH and SM showed greater diversity
than NR. For fungi, alpha diversity was more sensitive to pH and site
age, with lower diversity observed in alkaline and older soils.
In the plots below, we examine how the Shannon index, Faith PD and Evenness vary across different environmental and experimental conditions for the 16S (bacteria).
The boxplot below illustrate differences in Shannon diversity across groups. This metric reflects both species richness and how balanced the community is in terms of species abundance.
Here is a link to the full QIIME2 results (16S) : Shannon Index (16S)
Shannon entropy revealed significant overall differences, with green hay/bush sites showing higher diversity than both natural regeneration and seed mix sites. Seed mix sites also had higher Shannon diversity than natural regeneration.
Kruskal-Wallis p-value: 0.00102
The following plots show Faith’s Phylogenetic Diversity, which integrates evolutionary relationships to capture how phylogenetically broad each microbial community is.
Here is a link to the full QIIME2 results (16S) : Faith PD (16S)
Pielou’s evenness differed significantly across establishment methods (p = 0.0148), with green hay/bush sites being significantly more even than natural regeneration (q = 0.0148). The difference between seed mix and natural regeneration approached significance (q = 0.0597), while no significant difference was found between green hay/bush and seed mix.
Kruskal-Wallis p-value: 0.0106
These boxplots display Pielou’s Evenness, highlighting how uniformly taxa are represented in each community. It allows us to isolate imbalance in dominance from richness effects.
Here is a link to the full QIIME2 results (16S) : Pielou Evenness (16S)
Faith’s PD also varied significantly by establishment method (p = 0.0106), with both green hay/bush (q = 0.0148) and seed mix sites (q = 0.043) exhibiting greater phylogenetic diversity than natural regeneration. No significant difference was observed between green hay/bush and seed mix.
Kruskal-Wallis p-value: 0.0148
For the ITS dataset, the alpha diversity was done per sample. In order to align it with the 16S analysis—where samples were already grouped by site—we aggregated the alpha diversity values by computing the mean per site. Categorical metadata was simplified using the most common (modal) value per site. This ensures consistency across datasets in the visual outputs. However, users interested in the original, unaggregated sample-level data can explore the full QIIME 2 results via the links provided under each section.
For the variable pH_category, the median is 7.94 and for Age_category it’s 14 years.
The boxplot below illustrate differences in Shannon diversity across groups. This metric reflects both species richness and how balanced the community is in terms of species abundance.
Here is a link to the full QIIME2 results (ITS) : Shannon Index (ITS)
Fungal alpha diversity did not differ significantly between establishment methods or management types. However, Shannon diversity was significantly lower in alkaline soils (pH > 7.9, p < 0.02) and significantly higher in older sites (age > 14 years, p < 0.02), indicating that both soil pH and site age influence fungal richness and evenness.
Kruskal-Wallis p-value: 0.796
These boxplots display Pielou’s Evenness, highlighting how uniformly taxa are represented in each community. It allows us to isolate imbalance in dominance from richness effects.
Here is a link to the full QIIME2 results (ITS) : Pielou Evenness (ITS)
Sheep grazing was the only factor significantly affecting fungal evenness (p = 0.03), with grazed sites showing slightly higher Pielou’s Evenness. No significant effects were observed for establishment method, site age, or soil pH.
Kruskal-Wallis p-value: 0.512
To explore differences in microbial communities, we often rely on dimensionality reduction techniques such as Principal Coordinates Analysis (PCoA), visualised through Emperor plots. Two commonly used distance metrics in this context are Bray-Curtis and Jaccard.
While both metrics can reveal meaningful clustering and separation in microbial data, they capture complementary aspects of community structure.
Summary
Beta diversity patterns confirmed that establishment type, pH, and
grazing significantly shaped microbial community composition. For
bacteria, strong clustering by method and significant PERMANOVA results
support distinct assemblages. Fungal beta diversity followed similar
trends but with additional sensitivity to site age.
The Bray-Curtis Emperor plot is a 3D visualisation of microbial community dissimilarities between samples, based on the Bray-Curtis distance. This distance metric quantifies how different two samples are in terms of species abundance, taking into account both presence/absence and relative abundances. It does not incorporate evolutionary relationships between features.
Using Principal Coordinates Analysis (PCoA), the high-dimensional Bray-Curtis distance matrix is projected into a lower-dimensional space—typically three axes—to capture the main patterns of variation across samples.
The Emperor plot is an interactive 3D tool developed for QIIME 2 that allows users to explore these PCoA results. Samples are represented as points in space, and their spatial proximity reflects ecological similarity:
This type of plot is particularly useful for identifying clustering by experimental or environmental factors—such as establishment type, land management practices, or age group—and for detecting gradients or patterns in microbial community composition.
Bray-Curtis beta diversity analysis identified establishment method, pH, and sheep grazing as the strongest drivers of differences in bacterial community composition across sites. Site age also showed a significant effect with this metric.
PCoA plots based on Bray-Curtis distances revealed clustering of samples according to establishment method, with visible separation between natural regeneration, green hay/bush, and seed mix groups, although some overlap was observed.
Here is a link to the Bray-Curtis Emperor Plot for more flexibility on QIIME2: Bray-Curtis Emperor Plot (16S)
Bray-Curtis Emperor Plot
Bray-Curtis analysis identified establishment method, soil pH, and site age as the primary drivers of fungal community differences, followed by sheep grazing, ploughing, and cutting. In PCoA plots, natural regeneration sites formed a loose cluster near seed mix sites, while green hay/bush sites clustered more distinctly along Axis 1.
Here is a link to the Bray-Curtis Emperor Plot for more flexibility on QIIME2: Bray-Curtis Emperor Plot (ITS)
Bray-Curtis Emperor Plot (ITS)
The Bray–Curtis distance takes into account the relative abundance of taxonomic features, making it suitable for detecting quantitative differences in microbial communities across groups. When used in a PERMANOVA, this metric tests whether the overall structure and abundance of taxa vary significantly according to explanatory variables such as establishment type, age of the grassland, year of establishment, or management practices including cutting, ploughing, and the presence of grazing animals.. This approach highlights shifts in dominant or highly represented taxa.
Bray-Curtis-based PERMANOVA revealed significant differences in bacterial community composition across establishment methods, with all pairwise comparisons between methods also significant. These results indicate that restoration strategy has a strong influence on both the composition and abundance structure of bacterial communities.
Site age also had a significant but weaker effect. However, clustering by age in Bray-Curtis PCoA plots was less pronounced, suggesting more subtle shifts in community structure over time.
Soil pH significantly shaped bacterial community composition, with Bray-Curtis PCoA plots showing clear clustering between high and low-to-neutral pH soils
You can click on the images below to access the full QIIME2 report.
View full QIIME2 results (Bray–Curtis – Establishment)
Figure: PERMANOVA Bray–Curtis for “Establishment” (16S)
PERMANOVA analysis identified pH category (p = 0.001) and establishment method (p = 0.002) as the strongest drivers of differences in fungal communities. Site age (p = 0.024) and sheep grazing (p = 0.025) also had significant effects.
Pairwise comparisons showed the strongest differences between green hay/bush and seed mix sites (p = 0.008), followed by green hay/bush vs. natural regeneration (p = 0.011). Natural regeneration sites exhibited the highest variation in community composition, with Bray-Curtis distances ranging from 0.25 to 0.95.
You can click on the images below to access the full QIIME2 report.
View full QIIME2 results (Bray–Curtis – Establishment)
Figure: PERMANOVA Bray–Curtis for “Establishment” (ITS)
The Jaccard Emperor plot provides a 3D visualisation of microbial community dissimilarities based on the Jaccard distance. Unlike Bray-Curtis, the Jaccard metric considers only the presence or absence of features (e.g., microbial taxa) in each sample, ignoring their relative abundances.
This makes the Jaccard distance particularly suited for assessing community membership rather than abundance structure—focusing on which species are present, regardless of how abundant they are.
Using Principal Coordinates Analysis (PCoA), the high-dimensional Jaccard distance matrix is projected into a lower-dimensional space—usually three principal axes—to reveal major patterns in sample composition.
As with Bray-Curtis, the Emperor plot allows for interactive exploration of these ordinations:
The Jaccard plot is useful when exploring factors that influence community membership, such as habitat type, land use, or environmental filtering—especially in studies where presence/absence patterns are more meaningful than relative abundances.
Here is a link to the Jaccard Emperor Plot for more flexibility on QIIME2: Jaccard Emperor Plot (16S)
Jaccard Emperor Plot (16S)
Jaccard analysis confirmed similar drivers of community structure—establishment method, pH, and age—with natural regeneration, seed mix, and green hay/bush sites forming distinguishable but overlapping clusters in higher dimensions. This suggests that both presence/absence and abundance-based differences in fungal communities are shaped by restoration strategy and site conditions.
Here is a link to the Jaccard Emperor Plot for more flexibility on QIIME2: Jaccard Emperor Plot (ITS)
Jaccard Emperor Plot (ITS)
The Jaccard distance considers only the presence or absence of features, offering a more qualitative perspective on microbial community differences. In this PERMANOVA, the focus is on whether the composition of taxa — regardless of their abundance — differs significantly between groups. This metric is particularly useful for identifying patterns in taxon occurrence, including rare or transient species that may not strongly influence abundance-based distances.
You can click on the images below to access the full QIIME2 report.
View full QIIME2 results (Jaccard – Establishment)
Figure: PERMANOVA Jaccard for “Establishment” (16S)
With Jaccard dissimilarity, establishment method (p = 0.001) and pH (p = 0.002) remained the strongest drivers, followed by site age (p = 0.006), ploughing (p = 0.024), and cutting (p = 0.04).
Pairwise comparisons showed the most significant differences between green hay/bush and both natural regeneration and seed mix sites (p = 0.001), and also between natural regeneration and seed mix (p = 0.004). Median Jaccard dissimilarity values were high across all establishment methods (0.75–0.85), indicating substantial variation in taxon presence/absence.
You can click on the images below to access the full QIIME2 report.
View full QIIME2 results (Jaccard – Establishment)
Figure: PERMANOVA Jaccard for “Establishment” (ITS)
Here is a link to the Unweighted Unifrac Emperor Plot for more flexibility on QIIME2: Unweighted Unifrac Emperor Plot (16S)
Unweighted Unifrac Emperor Plot (16S)
You can click on the images below to access the full QIIME2 report.
View full QIIME2 results (Unweighted Unifrac – Establishment)
Figure: PERMANOVA Unweighted Unifrac for “Establishment” (16S)
Here is a link to the Weighted Unifrac Emperor Plot for more flexibility on QIIME2: Weighted Unifrac Emperor Plot (16S)
Weighted Unifrac Emperor Plot (16S)
You can click on the images below to access the full QIIME2 report.
View full QIIME2 results (Weighted Unifrac – Establishment)
Figure: PERMANOVA Weighted Unifrac for “Establishment” (16S)
Summary
Bacterial communities were dominated by Actinobacteria and
Proteobacteria, with variation by establishment and pH. Verrucomicrobia
was more frequent in NR sites. For fungi, Ascomycota, Mortierellomycota,
and Basidiomycota dominated, though many taxa remained unclassified.
Guilds such as saprotrophs, pathogens, and parasites varied with site
conditions.
Microbial community composition varied notably across establishment methods :
Green hay/bush soils were dominated by Alphaproteobacteria, Actinobacteria, and Bacilli
Natural regeneration sites had higher abundances of Acidobacteria-6 and Thermoleophilia, suggesting adaptation to low-nutrient conditions
Seed mix sites showed increased proportions of Alphaproteobacteria, Gammaproteobacteria, and Betaproteobacteria, indicating more copiotrophic (resource-demanding) communities typical of early-successional stages
Here is a link to the Taxonomy Barplots for more flexibility on QIIME2: Taxonomy Barplot (16S)
Taxonomy Barplots Associated with 'Establishment' (16S)
Here is a link to the Taxonomy Barplots for more flexibility on QIIME2: Taxonomy Barplot (ITS)
Taxonomy Barplots Associated with 'Establishment' (ITS)
To explore the composition of soil microbial communities, we used Krona plots — interactive, circular charts that display taxonomic abundances in a hierarchical manner.
These plots allow users to intuitively navigate from broader taxonomic levels (such as Phylum) to more specific ones (like Genus), while simultaneously comparing relative abundances across taxa.
In this study, Krona plots provide a powerful and user-friendly way to:
You can click on the images below to access the Krona plots for each site.
We used ANCOM to identify taxa whose relative abundances significantly differed across groups. This method accounts for the compositional nature of microbiome data by comparing log-ratios between taxa. The results are shown as volcano-like plots, where the W statistic reflects how many pairwise comparisons a taxon was found to differ in. Significant taxa are highlighted accordingly.
ANCOM analysis revealed that while most bacterial taxa did not vary significantly across establishment methods, a subset—including Intrasporangiaceae and Blastococcus (both Actinobacteria)—showed strong differential abundance (W > 300, clr > 15). These taxa responded notably to restoration strategy, indicating that establishment method can drive specific shifts in bacterial community structure.
Similarly, the ANCOM volcano plot for pH showed that although most bacterial taxa remained stable across the pH gradient, several taxa were strongly pH-responsive (W > 200). These results suggest that soil pH exerts a selective influence, favouring distinct bacterial groups in either more acidic or alkaline conditions.
View interactive volcano-like plot in QIIME2 (16S - Establishment )
Differences in Taxa Abundance Associated with 'Establishment' (ANCOM Results) (16S)
ANCOM analysis identified several fungal taxa as differentially abundant in relation to soil pH, site age, and establishment method:
Metapochonia (W = 386) was significantly more abundant in acidic soils, suggesting a preference for low-pH environments.
Two taxa were age-responsive: a Lasiosphaeriaceae taxon (W = 366) was more abundant in younger soils, while Paraphaeosphaeria (W = 320) was enriched in older soils, indicating shifts in fungal community structure with site maturation.
Regarding establishment method, Gibellulopsis (W = 596) was predominantly found in green hay/bush and seed mix sites, but rare in natural regeneration sites. In contrast, Metarhizium (W = 535) was most abundant in naturally regenerating sites, likely reflecting adaptation to less-disturbed, self-organising ecosystems.
These findings highlight how restoration strategy, pH, and time since restoration can drive specific taxonomic responses, shaping fungal community composition in grassland soils.
View interactive volcano-like plot in QIIME2 (ITS - Establishment )
Differences in Taxa Abundance Associated with 'Establishment' (ANCOM Results) (ITS)
Functional diversity refers to the variety of biological functions present in microbial communities. Unlike taxonomic analysis, which identifies who is there, functional analysis explores what they can do. This section includes two parallel analyses:
Alpha and beta diversity were analysed, along with differentially abundant functions or pathways.
Summary
Functional diversity revealed strong effects of establishment method and
pH on microbial pathway composition. Core metabolic functions were
consistent across sites, but specific biosynthetic and degradation
pathways differed by age, pH, and grazing. Saprotrophic functions were
dominant in fungi, with guild shifts linked to disturbance and soil
chemistry.
Summary
Bacterial functional alpha and beta diversity varied significantly
across establishment types, with GH and SM plots showing enhanced
diversity. Functional profiles also responded to pH and age. Several
metabolic pathways, including heme and amino acid biosynthesis, were
enriched in specific conditions, reflecting adaptive responses to
restoration.
The following metrics are based on PICRUSt2-inferred pathways.
We examined three complementary metrics: Shannon diversity (richness and evenness), Observed Features (richness only), and Evenness (dominance balance). You can use the menu below to explore how each metric varies according to different experimental variables.
Here is a link to the Shannon Index Boxplot for more flexibility on QIIME2: Shannon Index Boxplot - Functional (16S)
Kruskal-Wallis p-value: 0.00164
Here is a link to the Observed features Boxplot for more flexibility on QIIME2: Observed Features Boxplot - Functional (16S)
Kruskal-Wallis p-value: 0.0354
Here is a link to the Evenness Boxplot for more flexibility on QIIME2: Evenness Boxplot - Functional (16S)
Kruskal-Wallis p-value: 0.0294
Based on predicted pathway abundance from PICRUSt2.
Here is a link to the Bray-Curtis Emperor Plot for more flexibility on QIIME2: Bray-Curtis Emperor Plot (16S)
Bray-Curtis Emperor Plot (Functional)
View full QIIME2 results (Bray–Curtis – Establishment)
Figure: PERMANOVA Bray-Curtis for “Establishment” (ITS)
To identify pathways with significantly different abundances between groups, ANCOM was applied to PICRUSt2 pathway predictions.
Differences in Pathway Abundance Associated with 'Establishment' (ANCOM Results) (ITS)
These heatmaps display the relative abundance of predicted functional pathways across different samples, based on PICRUSt2-inferred metagenomic profiles from 16S rRNA gene data. Generated using QIIME2, each row represents a metabolic pathway (from MetaCyc), and each column corresponds to a sample group. The colour intensity indicates the predicted abundance of each functional pathway in the respective group.
You can click on each image to view a larger version in a new tab.
Summary
Fungal functional diversity was primarily influenced by pH and
establishment method. Lower functional diversity was observed in
alkaline soils. Guild analysis showed dominance of undefined
saprotrophs, with dung-associated and pathogenic guilds more prevalent
in managed or disturbed sites.
For fungal communities (ITS), functional diversity was assessed using metrics derived from predicted ecological functions (e.g., trophic modes, symbiosis potential).
The metrics below—Shannon diversity and Evenness—summarise the variability of ecological roles across different experimental conditions. Use the dropdown menu to explore patterns by variable.
Here is a link to the Shannon Index Boxplot for more flexibility on QIIME2: Shannon Index Boxplot - Functional (ITS)
Kruskal-Wallis p-value: 0.529
Here is a link to the Evenness Boxplot for more flexibility on QIIME2: Evenness Boxplot - Functional (ITS)
Kruskal-Wallis p-value: 0.226
Here is a link to the Bray-Curtis Emperor Plot for more flexibility on QIIME2: Bray-Curtis Emperor Plot (16S)
Bray-Curtis Emperor Plot (Fungi - ITS, Functional)
View full QIIME2 results (Bray–Curtis – Establishment)
Figure: PERMANOVA Bray-Curtis for “Establishment” (ITS)
Here is a link to the Jaccard Emperor Plot for more flexibility on QIIME2: Bray-Curtis Emperor Plot (16S)
Jaccard Emperor Plot (Fungi - ITS, Functional)
You can click on the images below to access the full QIIME2 report.
View full QIIME2 results (Jaccard – Establishment)
Figure: PERMANOVA Jaccard for “Establishment” (ITS)
In microbial ecology, a guild refers to a group of organisms that fulfil similar ecological roles, regardless of their taxonomic identity. Understanding functional guilds allows researchers to move beyond taxonomic profiles and assess the ecological roles that microbial communities may play in an environment.
To investigate the ecological roles of fungal communities, we used FUNGuild, a tool that assigns fungi to functional guilds based on curated databases and literature. These guilds represent ecological strategies such as:
This functional classification provides valuable insights into what fungi are likely doing in the ecosystem, beyond simply who they are.
This section explores the functional roles of fungi within each site, based on guild-level annotations provided by FUNGuild. Fungal guilds reflect ecological functions such as saprotrophy, symbiosis (e.g., mycorrhizal fungi), or pathogenicity. This approach provides insight into how fungal communities may contribute to ecosystem processes, complementing traditional taxonomic analyses.
The plot below highlights the top 20 most abundant fungal guilds identified using FUNGuild. To avoid clutter, the guild names are hidden on the y-axis; however, users can hover over each bar to reveal the full name, enabling interactive and detailed exploration of fungal functional diversity.

Top 20 functional guilds
The figure below shows the total abundance of fungal ASVs across sites, aggregated by functional guild. This provides an overview of how guild-level composition varies between locations, which may reflect differences in land use, soil conditions, or restoration histories.
Fungal communities were dominated by Undefined Saprotrophs, Plant Pathogens, and Dung Saprotrophs. A high proportion of Unassigned fungi highlights limits in current taxonomic resolution.
Fungal Guild Abundance by Site
Natural regeneration had the highest proportion of Undefined Saprotrophs, while Green Hay/Bush and Seed Mix had more Plant Pathogens and Dung Saprotrophs.
Alkaline sites showed dominance of Undefined Saprotrophs; acidic sites had more Animal Parasites.
Older sites had more Undefined Saprotrophs; younger sites showed higher proportions of Plant Pathogens and Dung Saprotrophs.
Fungal Guild Relative Abundance by Establishment
Summary
Correlations between bacterial and fungal taxa and functions revealed
potential ecological links, particularly between saprotrophic/pathogenic
fungi and sugar-degrading or nitrogen-associated bacterial pathways.
Most associations were site-specific and stronger in NR plots, with
nutrient cycling emerging as a shared functional axis.
Summary
Microbial diversity patterns strongly support the role of restoration
strategy and environmental variables in shaping both community
composition and function. Bacterial alpha diversity responded mainly to
establishment method, while beta diversity was influenced by pH,
grazing, and site age. Functional profiles mirrored these patterns, with
restoration type being the strongest driver. Despite these shifts, core
microbial functions remained stable, indicating a conserved metabolic
backbone across conditions.
Summary
Correlation analyses were limited by the complexity and abundance of
fungal-bacterial interactions. Filtering was required to manage the
volume of pairwise comparisons, which may have excluded meaningful
associations. The ambiguity of FUNGuild annotations also constrained
ecological interpretation. Alternative modelling approaches like GLMs
may yield more robust insights into cross-kingdom interactions.